Early Pattern of Differentiation in the
Human Pancreas
Michel Polak, Linda Bouchareb-Banaei, Raphael Scharfmann, and Paul Czernichow
In the early human embryonic/fetal pancreas, we studied 1) the ontogenetic pattern of the endocrine cells and
the evolution of the endocrine mass, and 2) the morphogenetic pattern of development and, more precisely,
the complex relationship of the epithelial mass with
the surrounding mesenchyme. We studied 15 pancreases between 7 and 11 weeks of development (WD) by
double immunohistochemistry. Epithelial cells in these
pancreatic anlage were detected by cytokeratin staining, and differentiated endocrine cells were detected by
insulin, glucagon, somatostatin, and pancreatic polypeptide staining. Proliferation was quantified using a
nuclear marker, the Ki-67 antibody. At this early stage,
the pancreas is made up of an epithelial mass composed
of central ducts intermingled with a loose mesenchyme
and peripheral ducts surrounded by a dense peripancreatic mesenchyme. Hormone-containing cells appear
in the epithelium at 8 WD. Newly differentiated endocrine cells coexpress insulin, glucagon, and somatostatin; endocrine differentiation starts within the central ducts of the epithelial mass, at a distance from the
dense peripancreatic surrounding mesenchyme. The
fraction of the primitive endocrine cells undergoing
proliferation is low (5% of the insulin cells at 8 WD, 3%
at 11 WD), which is in favor of massive differentiation
as the major mechanism for increasing endocrine mass.
By contrast, the nonendocrine epithelial cells have a
higher rate of proliferation; the epithelial cells in contact with the dense peripancreatic surrounding mesenchyme show more proliferation activity than those
within the central part of the epithelial mass (at 11
WD, labeling index: periphery 65% vs. center 15%, P <
0.001). In conclusion, the patterns of endocrine differentiation and epithelial proliferation observed within
the human pancreas early in development suggest that
the mesenchyme plays a role in these phenomena.
Diabetes 49:225–232, 2000
From the Institut National de la Santé et de la Recherche Médicale (INSERM)
U457 (M.P., L.B.-B., R.S., P.C.), the Department of Pediatric Endocrinology and
Diabetes (M.P., P.C.), Hôpital Robert Debré, Paris, France.
Address correspondence and reprint requests to Michel Polak, MD,
PhD, Department of Pediatric Endocrinology and Diabetes, INSERM U457,
48 Boulevard Sérurier, Hôpital Robert Debré, 75019 Paris, France. E-mail:
[email protected].
Received for publication 17 May 1999 and accepted in revised form
18 October 1999.
F G F, fibroblast growth factor; PBS, phosphate-buffered saline; WD,
weeks of development.
DIABETES, VOL. 49, FEBRUARY 2000
A
s is the case for other endodermal organs, the
development of the pancreas is thought to result
from interactions between the epithelium and
its associated mesenchyme. In murine models,
this has long been suspected (1). It has been shown in vivo,
in rodent models, that the mesenchyme is required for
proper differentiation of the primitive epithelium into
exocrine pancreas (2). It has also been shown, in an in vitro
rat model, that both the inductive effect of the mesenchyme
on the proper development of the exocrine tissue and its
repressive effect on the development of the endocrine cells
are mediated by soluble factors (3). However, the events
underlying the early period of pancreas development have not
yet been fully elucidated in humans.
The pancreas develops from the fusion of distinct endoderm-derived dorsal and ventral diverticula (4). In humans,
by day 35 of development, the ventral pancreatic bud begins
to migrate backwards and comes into contact and eventually
fuses with the dorsal pancreatic bud during the sixth week of
development (5,6). The dorsal bud gives rise to the major part
of the head, the body, and the tail of the future pancreas,
whereas the ventral bud gives rise to the inferior part of the
head of the future pancreas. Over the last 30 years there have
been a number of studies describing the ontogenetic pattern
of endocrine cells in the human pancreas (7–13). Indeed, a
description of the early appearance and development of cells
containing the four different hormones has been performed
from 8 to 40 weeks of gestation (9,10). More recently, differentiated human islet cells were shown to have a very limited
capacity for proliferation from 12 to 41 weeks of gestation
(14). This very low mitotic activity of the islet cells has been
advanced as an argument that the main increase of the islet
cell mass during fetal life in the developing pancreas is
brought about by islet cell neogenesis from hormone negative
precursor cells (14). However, in the existing data, the early
period from 7 to 11 weeks of gestation is not well described,
partly because of the scarcity of the material and partly
because of the imprecise determination of gestational age.
This imprecision is in sharp contrast with the recent identification of several transcription factors thought to control
early pancreatic development (15). The role of one of these,
PDX1, seemed to be confirmed when a patient with a pancreatic aplasia was shown to harbor a deleterious mutation
in the PDX1 gene (16).
Early pancreatic development is a crucial period during
which the primitive pancreas progresses from hormone negativity to synthesis of all the hormones later to be secreted by
the pancreas. We therefore undertook a study of the early
225
HUMAN PANCREAS EARLYDIFFERENTIATION
human fetal pancreas between weeks 7 and 11 of development 1) to describe the ontogenetic pattern of the four
endocrine cell types and the evolution of the endocrine mass
during this period, and 2) to describe the morphogenetic
development pattern during the same period and, more precisely, the complex relationship between the epithelial mass
and the surrounding mesenchyme.
RESEARCH DESIGN AND METHODS
Tissue collection and preparation. Human pancreases (n = 15) were dissected from embryonic and fetal tissue fragments obtained immediately after abortion performed between 6 and 11 weeks of development, as is permitted by
French law. The embryonic and fetal tissue collection and the experiments were
conducted according to the guidelines and with the approval of a committee for
protection of the person in biomedical research. Because a precise estimate of the
fetal age is of utmost importance at this early stage, gestational age was determined
from several developmental criteria: postovulatory age, as estimated from crownrump length measured during an ultrasound scan (17,18); hand and feet morphology, as compared with those described in an atlas of embryology (5,6,17); and
foot length (19). Up to 9 weeks of development (WD), these data were concordant, as has been previously described (17). Afterwards, in cases of discordant
data, we used the biparietal diameter and the crown-rump length from early
echographic measurements, because crown-rump lengths determined ultrasonically in vivo and in utero in cases of known postovulatory age (33–86 days)
agree well with those in length/age tables (17).
Abortions were induced mechanically, resulting in a warm ischemia time of
15–20 min. The tissue fragments were microdissected to allow removal of the
whole pancreas. No infectious disease or pancreatic pathologies were present in
the mothers. Whole pancreases were fixed in 3.7% buffered formalin during at least
6 h, then they were embedded in paraffin. Altogether, 15 pancreases were studied (three at 7 WD, four at 8 WD, three at 9 WD, two at 10 WD, two at 10.5 WD,
and one at 11 WD).
Immunohistochemistry
Sectioning of the pancreas. The whole pancreas was cut exhaustively into successive frontal sections from the dorsal to the ventral part. Paraffin sections (6 µm)
were mounted onto poly-L-lysine–coated slides and dried overnight at 37°C. For every
18 sections, throughout each pancreas, 6 consecutive sections were selected for
immunohistochemistry. This represented 6–10 studied sections for each hormone
and for a pancreas from 8 to 11 WD. The surface of the studied sections increased
with the increasing developmental age of the pancreas, due to the growth of the
organ, which occurred mainly along the frontal rather than the sagittal axis. At 7 and
8 WD, the pancreases were exhaustively cut, processed, and analyzed.
Double staining on paraffin-embedded sections. After deparaffinization and
rehydration, the slides were incubated three times for 4 min at a power of 900 W
in a microwave oven, in 0.01 mol/l citrate buffer (pH 6.0). Sections were then
allowed to cool at room temperature and were washed. After inhibiting the
endogenous peroxides with phosphate-buffered saline (PBS) solution containing 0.3% H2O2 for 10 min, the slides were washed in sterile water then incubated
with the blocking solution for 15 min (Biogenex; Menarini Diagnostics, ChevillyLarue, France). As a first step, for the proliferation study, the slides were then
incubated for 1 h with a monoclonal antibody to Ki-67 (clone Mib 1) diluted 1/50
in the solvent (Biogenex) (Table 1). The monoclonal Mib 1 antibody reacts with
the Ki-67 nuclear antigen of proliferating cells, which is expressed during all
phases of the cell cycle except for G0 and the beginning of the S phase (20).
Microwave treatment allows retrieval of protein antigenicity in paraffin-embedded tissue (21). The streptavidin-biotin-peroxidase complex was used with
diaminobenzidine as chromogen, producing brown nuclear staining (Biogenex).
For the second step of the double staining procedure, a monoclonal antibody to
either insulin or glucagon, or a polyclonal antibody to either somatostatin, cytokeratin, synaptophysin, or pancreatic polypeptide was applied (Table 1). Alkaline
phosphatase with fast red was used for antibody detection. In one experiment,
as a first step, an antibody to vimentin (a mesenchymal cells marker) was
applied, and for the second step, the antibody to cytokeratin (an epithelial cells
marker) was used. The immunohistochemical staining was controlled by parallel staining using only the secondary reagents (no primary antibodies) and by parallel staining using only the chromogen (no primary antibodies or secondary
reagents). All the controls were negative. For Ki-67, a positive control was made
by staining human tumors or human intestine, both of which have numerous proliferating cells. For the endocrine markers, positive controls were made on sections of normal human newborn or adult pancreases.
Immunofluorescence staining. To detect the coexpression of more than one
hormone in any cell type, pancreas sections were incubated 2 h with either
monoclonal anti-insulin and polyclonal anti-somatostatin, monoclonal antiglucagon and polyclonal anti-insulin, or monoclonal anti-glucagon and polyclonal
anti-somatostatin antibodies (Table 1). After being washed in PBS solution with
Tween 0.1%, the slides were incubated for 30 min with the second antibodies
(immunoglobulin G anti-mouse coupled with Texas Red [1/200], or anti-rabbit
[1/200] or anti–guinea pig [1/500] coupled with fluorescent detection system fluorescein isothiocyanate) and examined under the fluorescent microscope. Controls were performed by omitting the primary antibodies and by working with a
dilution of the anti-mouse coupled with Texas Red that did not cross-react with
the primary monoclonal antibodies raised in the guinea pig.
Quantitative analysis
Proliferating cells. The percentage of proliferating endocrine cells (labeling
index) was calculated for each hormonal cell type (insulin, glucagon, somatostatin,
pancreatic polypeptide) and for synaptophysin-containing cells as follows:
Hormone (or synaptophysin)-expressing cells:
Hormone (or synaptophysin) and Ki-67 double-positive cells
Total hormone (or synaptophysin)-positive cells
To determine the percentage of proliferating cells, 3,000–7,000 cells of each type
were counted per pancreas between 9 and 11 WD. A total of 40 insulin-expressing cells, 12 glucagon-expressing cells, and 21 somatostatin-expressing cells were
counted in the four pancreases at 8 WD.
To determine the percentage of proliferating duct cells, ducts were selected
either in the peripheral or in the central part of two pancreases of the same
developmental age. A total of 500 (7 WD) to 1,500 cells were counted per pancreas
and for each region (peripheral and central). The percentage of proliferating
duct cells was calculated as follows:
Cytokeratin-positive cells in the ducts and Ki-67–positive cells
Total cytokeratin-positive cells in the ducts
Hormonal coexpression. The percentage of cells expressing two hormones was
counted; the percentage of insulin cells coexpressing glucagon was expressed as
a ratio:
Glucagon-insulin double-positive cells
Total insulin-positive cells
TABLE 1
Primary antibodies used
Antigen
Insulin
Glucagon
Somatostatin
Pancreatic polypeptide
Synaptophysin
Cytokeratins
Vimentin
Ki-67
226
Antigen
species
Type
Antibody
species
Dilution
used
Laboratory
Dilution of the second and
third antibody (Biogenex)
Human
Porcine
Porcine
Human
Human
Human
Bovine
Human
Human
Monoclonal
Polyclonal
Monoclonal
Polyclonal
Polyclonal
Polyclonal
Polyclonal
Monoclonal
Monoclonal
Mouse
Guinea pig
Mouse
Rabbit
Rabbit
Rabbit
Rabbit
Mouse
Mouse
1/1,000
1/150
1/2,000
1/500
1/500
1/100
1/500
1/500
1/50
Sigma
Dako
Sigma
Dako
Biogenex
Dako
Dako
Dako
Immunotech
1/40
1/40
1/40
1/40
1/40
1/40
1/40
1/20
DIABETES, VOL. 49, FEBRUARY 2000
M. POLAK AND ASSOCIATES
We counted in a similar way the percentage of glucagon cells expressing insulin,
insulin cells expressing somatostatin, somatostatin cells expressing insulin,
glucagon cells expressing somatostatin, and somatostatin cells expressing glucagon.
The double fluorescent staining was carried out on three consecutive sections
of two pancreases of the same age, from the middle region of the pancreas. The
percentage was determined by counting 50–300 cells per pancreas.
Surface quantification. The insulin cell and epithelial cell surfaces (duct cells
expressing cytokeratin) were measured using a DMRB microscope (Leica Microsystems, Rueil-Malmaison, France) equipped with a color video camera connected to
a Quantimet 500 MC computer (screen magnification 10), as previously described
(22). All of the sections on which double immunostaining for insulin and Ki-67 had
been performed were quantified. The insulin and epithelial cell areas were expressed
in square micrometers or millimeters. The frontal sections from the dorsal part of
the pancreas (periphery of the organ) through the central part of the pancreas (center of the organ) to the ventral part (periphery of the organ) were quantified.
The insulin cell surface area was also normalized with regard to the epithelial
surface. The total number of insulin-containing cells was counted and normalized
per unit of epithelial surface to take into account the spatial heterogeneity of
endocrine cell location within the pancreas (see RESULTS).
RESULTS
Ontogenetic pattern of the endocrine cells
Chronological appearance of the endocrine cells. At
7 and 8 WD, the pancreas consisted of ducts lined by pyramidal cells (Fig. 1A and B). From 8 WD, the ducts were
embedded in a loose mesenchyme (Fig. 1B and C) and surrounded by a dense peripancreatic mesenchyme. At 9 WD, the
A
B
C
D
E
F
FIG. 1. At 7 (A) and 8 (B) WD, the pancreas consists of ducts lined by pyramidal cells, surrounded by a dense mesenchyme. At 8 WD,
immunoreactive insulin was detected (B, arrows). At 9 WD, the pancreatic ducts were branched and started to form a lobular pattern (C). The
lobular organization of branching ducts was clearly defined at 11 WD (D). Original magnification 100. The pancreas at this stage is composed
of epithelial cells and endocrine cells that are either isolated within the ducts (1), forming different-sized clusters in contact with the ducts
(2), or at a distance from the connective tissue (3) (D). Brown nuclear staining for Ki-67 identifies proliferating cells. E: Pancreas at 8 WD.
Insulin and glucagon double immunofluorescent staining is shown by arrows. All of the insulin-containing cells coexpress glucagon leading to
the orange staining. F: Pancreas at 11 WD: only a few insulin-containing cells (1, green fluorescence) coexpress glucagon (2, red fluorescence)
to give the orange staining (3).
DIABETES, VOL. 49, FEBRUARY 2000
227
HUMAN PANCREAS EARLYDIFFERENTIATION
A
B
FIG. 2. A: Double pancreatic immunostained cell at 10 WD. The red cytoplasm staining for the endocrine cell (here a somatostatin-containing
cell) surrounding the brown nucleus staining for Ki-67 (1) are the criteria retained for counting an endocrine cell as a proliferating cell. A
somatostatin-containing cell that is not proliferating is also shown (2). Original magnification 200. B: Labeling index of the insulin-,
glucagon-, somatostatin-, and synaptophysin-containing cells according to the developmental age of the human pancreas.
pancreatic ducts were branched and started to form a lobular pattern (Fig. 1C). The lobular organization of branching
ducts became apparent by 10 WD and was clearly defined at
11 WD (Fig. 1D). After exhaustive analyses of all sections of
the whole pancreas, no immunoreactive hormone was
detected at 7 WD (Fig. 1A). Between 8 and 11 WD, endocrine
cells appeared either as single cells or as cell clusters but did
not form islets of Langerhans. The first endocrine cells were
seen at 8 WD when immunoreactive insulin, glucagon, and
somatostatin were detected in three out of four of the pancreases tested (Fig. 1B). Pancreatic polypeptide was
detected from 9 WD and onward. At that stage, the hormonecontaining cells were found either in the epithelial structure
(ducts) or in cell groups in contact with or distant from the
ducts (Fig. 1D).
Hormonal coexpression. At 8 WD, 92% of cells marked
with the anti-insulin antibody were also marked with the antiglucagon antibody (Fig. 1E), and correspondingly, 97% of cells
marked with the anti-glucagon antibody were also marked
with the anti-insulin antibody. All cells marked with the anti228
insulin antibody were also marked with the anti-somatostatin
antibody, and correspondingly, 93% of cells marked with the
anti-somatostatin antibody were also marked with the antiinsulin antibody. All cells marked with the anti-glucagon antibody were also marked with the anti-somatostatin antibody,
and correspondingly, 92% of cells marked with the antisomatostatin antibody were also marked with the antiglucagon antibody. At 9 WD and thereafter, percentages of cells
coexpressing more than one hormone dropped dramatically:
11% of cells marked by the anti-insulin antibody were also
marked by the anti-glucagon antibody, and 10% of cells
marked with the anti-glucagon antibody were also marked
with the anti-insulin antibody (Fig. 1F); 6% of cells marked with
the anti-insulin antibody were also marked with the antisomatostatin antibody at 11 WD.
Endocrine cell proliferation. Endocrine cell density and
endocrine surface increased dramatically between 8 and 11
WD, from 20 insulin cells/mm2 of epithelial surface to more
than 600. Double immunocytochemical staining revealed that
the cell proliferation marker Ki-67 was detectable in nuclei
DIABETES, VOL. 49, FEBRUARY 2000
M. POLAK AND ASSOCIATES
FIG. 3. A: Human pancreas at 9 WD. At this early stage, the pancreas is
made up of an epithelial mass, composed of central ducts intermingled
with a loose mesenchyme and peripheral ducts surrounded by a dense
peripancreatic mesenchyme. Insulin-containing cells (arrows) were
found in the central epithelial structures surrounded by a rim of ducts
in which no endocrine differentiation was seen to occur. The endocrine
cells were therefore located at a distance from the dense peripancreatic surrounding mesenchyme. Original magnification 50. B: Human
pancreas at 9 WD. The mesenchyme was marked with an antibody to
vimentin (brown staining) and the duct cells with an antibody to cytokeratin (polarized red staining in the cytoplasm, arrows). The central
ducts are seen intermingled with a loose mesenchyme and the peripheral ducts are seen surrounded by a dense peripancreatic mesenchyme.
Original magnification 100. C: Human pancreas at 9 WD. Higher magnification of A. The endocrine cells can be seen to be dispersed in heterogeneous groups. The same insulin-containing cells (red, arrows) as
in A were seen in the central epithelial structures (central ducts) and
were surrounded by a rim of epithelial structures (peripheral ducts) in
which no endocrine differentiation had occurred. The central pattern
of endocrine localization is seen. Original magnification 100.
DIABETES, VOL. 49, FEBRUARY 2000
from insulin-, glucagon-, somatostatin-, and synaptophysin-positive cells (Fig. 2A for somatostatin-positive cells, not shown
for the other hormonal cell types). Synaptophysin-positive
nerve cells were recognized morphologically and were not
included in the counts. The percentage of hormone-expressing cells seen to be proliferating remained low from their
appearance at 8 WD to 11 WD. Indeed, 5% of the insulin cells
were Ki-67–positive at 8 WD (out of 40 insulin-containing cells
present in three out of four pancreases), and 3% at 11 WD (out
of 7,000 insulin-containing cells counted) (Fig. 2B).
Morphogenetic pattern of development in the early
human pancreas
Spatial heterogeneity of the endocrine cells in the
pancreas throughout early development. From 7 to 11
WD, the pancreas consists of an epithelial tube surrounded
by mesenchyme. At 8 WD and onward, the pancreas is made
up of an epithelial mass, composed of central ducts intermingled with a loose mesenchyme and peripheral ducts surrounded by a dense peripancreatic mesenchyme (Fig. 3A
and B). The endocrine cells can be seen to be dispersed in heterogeneous groups. Indeed, hormone-containing cells were
seen in the central epithelial structures (central ducts) and
were surrounded by a rim of epithelial structures (peripheral
ducts) in which no endocrine differentiation had occurred
(Fig. 3A and C).
As shown in Fig. 4, the dispersion of each cell type was
quantified in two ways. The absolute number of endocrine
cells, as measured by the surface of insulin-positive cells,
was higher in the center of the pancreatic organ than at the
periphery of the pancreatic organ (Fig. 4A and C; see sectioning of the pancreas and surface quantification in
RESEARCH DESIGN AND METHODS). Also when expressed as
insulin-positive cell surface per unit of epithelial surface, the
insulin-containing cell density increased from the periphery
to the center of the pancreatic organ. This allowed us to
exclude the possibility that a sampling bias might explain the
observed pattern (Fig. 4B and D).
The endocrine cells were located in the more central ductal epithelial structures sometime in contact with the loose
interductular mesenchyme but at a distance from the dense
peripancreatic surrounding mesenchyme. Identical patterns
were obtained for the other 10 insulin-containing pancreases
analyzed and quantified in the same way (Fig. 4, data shown
for two pancreases).
With increasing developmental age (from 8 to 11 WD), the
endocrine surface increased in the center of the pancreatic
organ and toward the more peripheral areas of the pancreatic
organ (Fig. 5). These findings are consistent with a centrifugal evolution of the endocrine mass in the early developing
pancreas.
Spatial heterogeneity in the capacity for proliferation
of the epithelial duct cells in the human pancreas
during early development. The proliferating cells in the
pancreatic ducts also appeared to be heterogeneously distributed. We therefore quantified the proliferating cells in the
ducts in close relation with the dense peripancreatic surrounding mesenchyme (peripheral ducts) and those in the
ducts located in the center of the tissue (central ducts). Proliferation of the peripheral ducts was much greater than that
of central ducts (65 vs. 15% Ki-67–labeled duct cells at 11 WD,
P < 0.001). The same phenomenon was observed between
7 and 11 WD (Fig. 6).
229
HUMAN PANCREAS EARLYDIFFERENTIATION
A
C
B
D
FIG. 4. A and C: Quantification of the insulin ( ) cell surface of a 9-WD (A) and an 11-WD (C) human pancreas. The 9-WD pancreas was exhaustively cut in frontal sections from section 1 (S1, dorsal) to section 200 (S200, ventral) in consecutive 6-µm thick sections. The 11-WD pancreas
was also exhaustively cut in frontal sections from section 1 (S1, dorsal) to section 250 (S250, ventral). The insulin cell and epithelial cell surfaces (duct cells expressing cytokeratin) were measured using a Leica DMRB microscope equipped with a color video camera connected to a
Quantimet 500 MC computer. A total of 10–12 regularly sampled frontal sections from the dorsal part of the pancreas (periphery of the organ)
through the central part of the pancreas (center of the organ) to the ventral part (periphery of the organ) were quantified. This shows the
central pattern of endocrine differentiation within the pancreas (A and C). On the same sections, the insulin surface was expressed per unit
of epithelial cell surface (B and D). It was much greater in the central area than in the peripheral areas of the pancreatic organ. See also sectioning of the pancreas and surface quantification in RESEARCH DESIGN AND METHODS.
DISCUSSION
In this study we examined early human pancreatic development in a window of time in which the pancreas evolves
from a hormone-negative to a hormone-expressing stage.
The above results concerning the apparition of each
endocrine cell type are concordant with certain studies
(9,10) but discordant with others, in which no glucagon-containing cells were detected at 8 WD or in which cells containing somatostatin and pancreatic polypeptide were
detected at 7 WD (11,13). The antibodies used, the sensitivity of the immunohistochemistry method, and also the dating
of the developmental age may account for these differences.
Concerning the dating method used here, up to 9 WD, the morphological data and in vivo ultrasonographical measurements were concordant and thus gave a very precise developmental age (5,6,17,18). This rigorous technique for dating
the fetus thus allows us to be confident that insulin-containing cells are already present at 8 WD. At this stage, almost all
the insulin cells coexpress glucagon and somatostatin. The
majority of insulin-containing cells stained negative for the
other pancreatic hormones after 9 WD, which is regarded as
a sign of maturation by many authors (22,24,25).
230
FIG. 5. Human pancreas at 11 WD. A central pattern of endocrine localization is seen. The endocrine surface (2) increased in the center of
the pancreas toward the more peripheral areas and toward the peripheral ducts of the pancreatic organ (1). Peripancreatic mesenchyme (M)
is also shown. Original magnification 50.
DIABETES, VOL. 49, FEBRUARY 2000
M. POLAK AND ASSOCIATES
A
B
FIG. 6. Proliferation of the pancreatic epithelial duct cells in the
early developing human pancreas. A: Human pancreas at 7 WD. The
brown nuclear staining for Ki-67 identifies proliferating cells. The
peripheral duct cells (arrowheads) seem to proliferate more than
central duct cells (arrows). Original magnification 100. B: Percentage of pancreatic duct cells undergoing proliferation during development. The labeling index of the peripheral duct cells ( ) and the central duct cells ( ) show a regional heterogeneity in the capacity of the
epithelial duct cells’ proliferation.
We used immunohistochemical staining of Ki-67 to detect
proliferating human fetal cells in vivo and for comparison
with previous data obtained from older human pancreases
at 12–41 WD (14). Ki-67 immunoreactivity is thought to give
an approximate estimate of the fraction of cells that are
cycling (20). The data obtained here showing low prolifer ation activity of the endocrine cells (5 and 3% of the insulincontaining cells at 8 and 11 WD, respectively) are consistent
with previously published data (14). Such weak proliferation
activity even during very early pancreatic development is in
line with previous in vitro studies on fetal and adult human
islet cells, in which endocrine cell proliferation was found
to be very low (24,26–28). These data, taken together with
the very large increase in the density and number of
endocrine cells found here (even if the precise cell cycle timing of early fetal endocrine cells is not known, precluding
exact calculation), are consistent with the hypothesis that
increased endocrine cell mass in the early developing
human pancreas results from massive differentiation of
unspecialized cells.
DIABETES, VOL. 49, FEBRUARY 2000
The spatial organization of endocrine differentiation
within the developing pancreas is a major finding of this
study. In all of the studied pancreases in which endocrine differentiation had occurred, the endocrine cells were located
in the central epithelial ducts of the pancreas separated from
the surrounding dense peripancreatic mesenchyme by
epithelial structures (ducts) that were free of endocrine differentiation. This was revealed by the systematic analysis of
very early developing pancreases, which show relatively simple tissue organization, i.e., an epithelial tube with little
branching and duct cells embedded in a loose mesenchyme,
surrounded by a dense peripancreatic mesenchyme. It has
been shown in a rat in vitro model that recapitulates embryonic pancreatic development that the mesenchyme has a
repressive effect on the development of the endocrine tissue
(3). The current study suggests that this could indeed be the
case in the early development of the human pancreas.
Moreover, we show here a very high proliferation ability of
the duct cells, particularly of those located in close vicinity
to the dense peripancreatic mesenchyme where the proliferation is two to three times more than in the central ducts
and constant from 7 to 11 WD. This suggests that the mesenchyme has an inductive effect on the proliferation of the
duct cells, and therefore it probably plays a role in the proper
growth and subsequent differentiation of the exocrine pancreas, which ultimately represents the greatest mass of the
pancreas. This is consistent with the mesenchyme effect
observed on the exocrine pancreas in a rat in vitro model (3).
Proliferation and differentiation of progenitor cells in the
developing murine embryonic anterior pituitary have also
been shown to respect a spatial organization (29). In the case
of the pituitary, this pattern was shown to be linked to spatial
and temporal restriction in fibroblast growth factor (FGF) and
bone morphogenetic protein-mediated signals from adjacent
neural and mesenchymal signaling centers (29). It is of interest to note that in the developing rat pancreas, FGF2 promotes pancreatic epithelial cell proliferation (30). In the rat
model, follistatin can mimic both inductive and repressive
effects of the mesenchyme. Follistatin might thus represent one
of the mesenchyme factors required for exocrine tissue development while exerting a repressive effect on endocrine cell differentiation (3). Whether this spatial duct and endocrine development pattern is linked only to soluble mesenchyme-secreted
factors or to differences in the intrinsic differentiation ability
of the epithelium related to location remains to be established.
In conclusion, we have shown that in the human pancreas
1) hormone-containing cells appear at 8 WD when they coexpress insulin, glucagon, and somatostatin; 2) the ability of the
primitive endocrine cells to proliferate is very limited,
thereby suggesting that the major mechanism behind the
increase in endocrine mass is differentiation from hormonenegative precursor cells; 3) there is a spatial and temporal pattern of endocrine differentiation that starts within the central
epithelial ducts at a distance from the dense peripancreatic
surrounding mesenchyme and shows a centrifugal evolution
from 8 to 11 WD; and 4) there is very intense duct proliferation in proximity to the surrounding mesenchyme: this suggests that both the inductive effect of the mesenchyme on the
development of the primitive epithelial tissue (ducts) and its
repressive effect on the development of the endocrine cells
are operating in the developing early human pancreas, as in
the previously described rodent model (3).
231
HUMAN PANCREAS EARLYDIFFERENTIATION
ACKNOWLEDGMENTS
This study was supported by a Juvenile Diabetes Foundation
grant (198207) to R.S. and by the Aide aux Jeunes Diabétiques,
with a grant to M.P. and an educational grant to L.B.-B.
We wish to thank Professor M. Peuchmaur and his team from
the pathology department at the Hôpital Robert Debré for
continuous help during this work. We are indebted to the medical staff in the department of gynecological surgery at the
Hôpital Robert Debré, directed by Professor Blot, and to Dr.
Benifla in the department of gynecological surgery at the Hôpital Bichat for providing human embryonic and fetal tissues.
REFERENCES
1. Golosow N, Grobstein C: Epithelio-mesenchymal interaction in pancreatic
morphogenesis. Dev Biol 4:242–255, 1962
2. Gittes GK, Galante PE, Hanahan D, Rutter WJ, Debas HT: Lineage-specific morphogenesis in the developing pancreas: role of mesenchymal factors. Devel opment 122:439–447, 1996
3. Miralles F, Czernichow P, Scharfmann R: Embryonic mesenchyme regulates
the relative proportions of endocrine versus exocrine tissue during pancreatic development. Development 125:1017–1024, 1998
4. Slack JMW: Developmental biology of the pancreas. Development 121:1569–
1580, 1995
5. Moore KL, Persaud TVN, Shiota K: Color Atlas of Clinical Embryology.
Philadelphia, WB Saunders, 1995, p. 153–163
6. O’Rahilly R, Müller F: Developmental Stages in Human Embryos. Meriden,
CT, Meriden-Stinehour Press, Carnegie Institution of Washington (publ. no.
637), 1987
7. Falin LI: The development and cytodifferentiation of the islets of Langerhans in human embryos and foetuses. Acta Anat 68:147–168, 1967
8. Like A, Orci L: Embryogenesis of the human pancreatic islets: a light and electron microscopic study. Diabetes 21 (Suppl. 2):511–534, 1972
9. Stefan Y, Grasso S, Perrelet A, Orci L: A quantitative immunofluorescent
study of the endocrine cell populations in the developing human pancreas. Dia betes 32:293–301, 1983
10. Clark A, Grant AM: Quantitative morphology of endocrine cells in human fetal
pancreas. Diabetologia 25:31–35, 1983
11. Fukayama M, Ogawa M, Hayashi Y: Development of human pancreas:
immunohistochemical study of fetal pancreatic secretory proteins. Differen tiation 31:127–133, 1986
12. Von Dorsche H, Fält K, Titlbach M: Immunohistochemical, morphometric, and
ultrastructural investigations of the early development of insulin, somatostatin,
glucagon, and PP cells in foetal human pancreas. Diabetes Res 12:51–56, 1989
13. Bocian-Sobkowska J, Zabel M, Wozniak W, Surdyk-Zasada J: Prenatal development of the human pancreatic islets: immunocytochemical identification
of insulin-, glucagon-, somatostatin-, and pancreatic polypeptide-containing
232
cells. Folia Histochem Cytobiol 35:151–154, 1997
14. Bouwens L, Lu WG, De Krijger R: Proliferation and differentiation in the
human fetal endocrine pancreas. Diabetologia 40:398–404, 1997
15. Edlund H: Transcribing pancreas. Diabetes 47:1817–1823, 1998
16. Stoffers DA, Zinkin NT, Stanojevic V, Clarke WL, Habener JF: Pancreatic agenesis attributable to a single nucleotide deletion in the human IPF1 gene
coding sequence. Nat Genet 15:106–110, 1997
17. Drumm JE, O’Rahilly R: The assessment of prenatal age from the crown-rump
length determined ultrasonically. Am J Anat 148:555–560, 1977
18. Guihard-Costa AM, Larroche JC: Fetal biometry. In Fetal Diagnosis and
Therapy. Basel, Karger, 1995, p. 212–278
19. Munsick R: Human fetal extremity lengths in the interval from 9 to 21 menstrual weeks of pregnancy. Am J Obstet Gynecol 8:883–886, 1984
20. Gerdes J, Lemke H, Baisch H, Wacker H, Schwab U, Stein H: Cell cycle analysis of a cell proliferating–associated human nuclear antigen defined by the
monoclonal antibody Ki-67. J Immunol 4:1710–1715, 1984
21. Cattoretti G, Becker M, Key G: Monoclonal antibodies against recombinant
parts of the Ki-67 antigen (MIB 1 and MIB 3) detect proliferating cells in
microwave-processed formalin-fixed paraffin sections. J Pathol 168:357–363,
1992
22. Garofano A, Czernichow P, Bréant B: In utero undernutrition impairs rat
-cell development. Diabetologia 40:1231–1234, 1997
23. Lukinius A, Ericsson JLE, Grimelius L, Korsgren O: Ultrastructural studies of
the ontogeny of fetal human and porcine endocrine pancreas, with special reference to colocalization of the four major islet hormones. Dev Biol 153:376–
385, 1992
24. De Krijger RR, Aanstoot HJ, Kranenburg G, Reinhard M, Visser WJ, Bruining
GJ: The midgestational human fetal pancreas contains cells coexpressing
islet hormones. Dev Biol 153:368–375, 1992
25. Miralles F, Serup P, Cluzeaud F, Vandewalle A, Czernichow P, Scharfmann R:
Characterization of -cells developed in vitro from rat embryonic pancreatic
epithelium. Dev Dyn 214:116–126, 1999
26. Otonkoski T, Beattie GM, Rubin JS, Lopez AD, Baird A, Hayek A: Hepatocyte
growth factor/scatter factor has insulinotropic activity in human fetal pancreatic cells. Diabetes 43:947–953, 1994
27. Hayek A, Beattie GM, Cirulli V, Lopez AD, Ricordi C, Rubin JS: Growth factor/matrix-induced proliferation of human adult -cells. Diabetes 44:1458–1460,
1995
28. Beattie GM, Rubin JS, Mally MI, Otonkoski T, Hayek A: Regulation of proliferation and differentiation of human fetal pancreatic islet cells by extracellular matrix, hepatocyte growth factor, and cell-cell contact. Diabetes 45:
1223–1228, 1996
29. Ericson J, Norlin S, Jessell TM, Edlund T: Integrated FGF and BMP signaling
controls the progression of progenitor cell differentiation and the emergence
of pattern in the embryonic anterior pituitary. Development 125:1005–1015, 1998
30. Le Bras S, Miralles F, Basmaciogullari A, Czernichow P, Scharfmann R: Fibroblast growth factor 2 promotes pancreatic epithelial cell proliferation via
functional fibroblast growth factor receptors during embryonic life. Diabetes
47:1236–1242, 1998
DIABETES, VOL. 49, FEBRUARY 2000